How to read the sky of a world you will never see

When a planet passes in front of its star, a tiny fraction of the starlight filters through the planet’s atmosphere on its way to us. Different molecules absorb different wavelengths, so at those wavelengths the planet appears very slightly larger. By comparing the star’s spectrum during and outside the transit, astronomers can work out what the atmosphere contains. The technique is called transmission spectroscopy, and the James Webb Space Telescope has made it routine.

K2-18b is its most argued-over target. The planet orbits a red dwarf star about 124 light-years away in the constellation Leo. It is about 2.6 times Earth’s radius and nearly nine times its mass, and it sits in the zone where a planet could, under the right conditions, hold liquid water.

How small the signal is

Some rough arithmetic shows what astronomers are up against. K2-18b blocks about 0.29 percent of its star’s light as it crosses, or 2,900 parts per million. The atmosphere’s contribution is a thin ring around that silhouette. Its thickness is set by the scale height, the distance over which pressure drops by a factor of about 2.7. For a warm, hydrogen-rich atmosphere under this planet’s gravity, that comes to around 75 kilometers.

A ring one scale height thick covers about 26 parts per million of the star. An absorbing molecule might raise the apparent atmosphere by a few scale heights at its favored wavelengths, so the entire feature amounts to a few tens of parts per million. Webb has to detect a change of about one part in 30,000 in the star’s brightness, at a chosen color, during a transit that lasts a few hours.

It also explains why this planet was picked. Run the same calculation for Earth crossing the Sun and one scale height comes to about 0.2 parts per million, more than 100 times smaller. A puffy hydrogen atmosphere around a small star is the easy case. Finding the signature of a true Earth twin is, for now, out of reach.

A claim and its unraveling

In 2023, Webb observations detected methane and carbon dioxide in K2-18b’s atmosphere, and a notable absence of ammonia. A group led by Nikku Madhusudhan argued that this combination fits a hycean world: a planet with a global ocean under a hydrogen-rich atmosphere. The same group reported a faint hint of dimethyl sulfide, a molecule that on Earth is produced mainly by marine life.

In 2025 they reported a stronger signal of dimethyl sulfide or a related molecule using a different Webb instrument, at about three standard deviations. The announcement drew wide attention and swift criticism. Several independent reanalyses of the same data found that the signal’s strength depended heavily on modeling choices.

Then more data arrived. A team led by Renyu Hu added further Webb transits and reported carbon dioxide at 3.7 standard deviations with no detection of dimethyl sulfide, interpreting the planet as having a water-rich interior. A study published in Astronomy & Astrophysics in 2026 found that the tentative dimethyl sulfide signal vanishes in its most comprehensive models, that even the carbon dioxide evidence is weak and model-dependent, and that the planet is more likely a volatile-poor world than a hycean one.

The signal’s strength depended heavily on modeling choices.

Why the same data give different answers

Nobody involved is being careless. Extracting a spectrum this faint involves dozens of decisions: how to remove instrument systematics, how to model the star, which molecules to allow, what range of temperatures is permitted. Red dwarfs are spotty and active, and their own variability can mimic or mask atmospheric features. A molecule like dimethyl sulfide shares absorption features with other carbon-bearing gases, so whether it is “detected” depends partly on which alternatives were offered to the fit.

The interpretation is looser still. K2-18b has been described as a hycean ocean world, a mini-Neptune with a deep, hot atmosphere, a planet with a magma ocean, and a water-rich body with no accessible surface. Each can be made to produce a similar small set of gases. In a sense the question of what the planet is was answered in advance by the prior assumptions of each study.

How to read the next headline

K2-18b will not be the last claim of its kind, and it offers a practical test for the next one. Ask how many transits went into the result, and whether a different team has processed the data. Ask whether the molecule is claimed outright, or only preferred over a list of chosen alternatives. Ask what the significance would be if the simplest competing model were used. And ask what the planet is assumed to be before any molecule is interpreted, since a biosignature is only meaningful in a place where you understand the non-biological chemistry.

None of this means the observations were wasted. The debate has sharpened statistical methods, prompted new laboratory measurements of molecular spectra, and made clear how much observing time a convincing detection would require. The search for life on other planets will almost certainly proceed this way: a hint, a challenge, a better measurement, and many rounds of argument. That is what the process looks like when it works, and the same pattern is playing out on Mars, with rocks instead of starlight.

Sources

• Madhusudhan et al., “Carbon-bearing Molecules in a Possible Hycean Atmosphere”, 2023
• Madhusudhan et al., “New Constraints on DMS and DMDS in the Atmosphere of K2-18 b from JWST MIRI”, 2025
• Hu et al., “A water-rich interior in the temperate sub-Neptune K2-18 b revealed by JWST”, 2025
• Fernández-Rodríguez et al., “The atmospheric composition of the sub-Neptune K2-18b and insights into its formation”, Astronomy & Astrophysics, 2026